Here’s the thing for me:

we can try hard now or just wait 6-12 months and not try hard at all!

I totally get the up skilling thing but….

Why bother when these learnings will be redundant shortly.

You probably right @Chagsy

For me it's less about the tech and more about getting my framework and process organised. I'm not a full time investor so keeping up with everything while working is genuinely challenging compared to someone who does this all day.

I'm a long term investor at heart, this just helps me turn rocks over quicker. Nothing I built is groundbreaking either! But it's worth it for where I'm at right now, and I'll just adapt it as things evolve. 

Thanks for posting @Magneto, always good to see what others are up to on this front.

The @JohnnyM video was insightful and is on my list to revisit.

Your sentiment makes sense @Chagsybut I’m taking the other side of that bet.

I am doubling down (at least trying to) on AI skills (literal and more general) to build some fluency in prep for what’s coming.

I’ve been in the situation many times in corporate life when a big system project comes through and the people who were there before it and during have a much deeper understanding about what’s going on behind the screen than those who came after.

Getting in now on the ground floor (or 3^rd, 4^th wherever we are up to now) when the backend infrastructure seems stable and the front end is not seamless but still being built out and pivoting on the way up should be a bit like building corporate memory I reckon. Difficult to value / put a price on but impossible to recreate after the event.

Plus the capabilities are sharp enough now to warrant the effort – for me at least.

Anyway, that’s my take, each to their own.

I’m in a similar headspace to you @Slomo . A bit of education and understanding now for an undefined payback later.

For me it’s taking the form of some basic tools being built into business workflows.

Hardly revolutionary but not being done by any of my competitors at the moment.

That being said, they will be. At best I’m feeling a 2 year runway until everything I’m doing now will be SOP but for now my SOP will be their ‘bleeding edge’ methodology they try and sell to clients that we share.

I’d also expect that the tools will become redundant and more built in features of standard software suites. Thus even less ‘bleeding edge’.

Well done @Magneto on challenging yourself on finding new ways to streamline your process.

I've a similar daily news scanner for my portfolio companies and it's a great way of keeping myself informed on the latest happenings. These types of workflows can be supercharged by augmenting them with your own notes with information that is not widely found on the internet. For example, customers you know the company has strong relationships with, potential contracts that are upcoming, some key events/situations you would like to keep an eye on. It'll provide the necessary hints and focus to the LLM to do more targeted searches and uncover news and relationships that you've not been aware of.

I've also found it useful to let the LLM keep a persistent log of the news its found, so it doesn't double up on notifying you the same things over and over again.

Thanks @mushroompanda this is exactly where I’m trying to head with it.

My main aim is to free up more time for deeper research on my stocks rather than just reacting to surface-level news. Longer term, I’d like to move toward being a full-time investor, and I think the real edge will come from doing that kind of work that isn’t easily found online, things like industry nuances, relationships, and forward-looking insights.

AI is definitely going to level the playing field when it comes to access to information, but I still think there’ll be a clear gap between winners and losers. The difference will be how well you can interpret that information and combine it with your own original thinking.

The biggest issue in my house right now is the fight for Tokens.. My 13yo is creating a program that generates Roblox Assets he can sell to his mates.. I’ve got subscriptions for ChatGPT, Claude and GitHub.. he uses them all but I’m not going to get him his own… he burns through my Tokens like they are McDonalds French fries…

Thanks @Magneto and everyone else, for the follow up on this post.

In sharing this update about Bob and his mates, Sally, Wally and a few others.

I’m sharing access to my GitHub Repo (Google it, I had to…) it’s a public Repo which can be found here so you can just take a copy and ask your AI engine to compare and contrast the coding to whatever you’ve created.. In the spirit of the forum I that hope helps… if yours is better than mine, feel free to share….

Also sharing the Claude Skill that I’ve created. Claude Skills seem to be all the rage.. this one essentially instructs Claude on how to behave like a buy side analyst, it’s an accumulation of a whole bunch of master prompts I’ve been refining over time. If you’re on the paid version of Claude you can just plug in the Skill instructions at the bottom of the post, the Skill itself can be downloaded from here. When installed just type “/ASX-Investor BHP” and it will spit out a deep dive on BHP. You can then ask Claude to enhance the Skill as you see fit. You can use it when HY or FY results are announced. Again if you coming up with something amazing.. feel free to ping it back.

If you stop reading here that’s cool, below is a slightly longer explanation and visualisations of what these Reporting Agents and Claude Skills do.

Cheers

JM

Bob has my list of holdings, he goes to the ASX website every morning at 9am (7am SG time) and looks for any announcements from the 24 companies I own shares in. If it’s something chunky he passes the PDFs to Claude to work his magic.

Wally is my Watchman.. he has all my watchlists included TII75 which is the Top 30 compounders of all time as identified by Jason Zweig in the 75^th anniversary edition of The Intelligent Investor. Happy to share a pic of those pages if you’ve not see the latest copy of the book, Resmed is included!! Wally has a Spreadsheet he produces which essentially gives me a 10yr view of Price, P/E Multiples and a Div/RoRR Valuation metric. I’m still working on the format of his output.. it’s a bit hit and miss.

Sally is my Selling bot, she looks at the same Portfolio as Bob but tries to convince me it’s time to sell.

Below is an attempt (by Claude) to visualise what’s going on here.

Below is an attempt to visualise what the Claude Skill ASX-Investor does in various situations.

They'll need Claude Pro or Team (paid plan). Then it's four steps:

1. Go to claude.ai, click Projects in the left sidebar, and create a new project — name it something like "ASX Investor."
2. Inside the project, click "Set project instructions" and paste the entire contents of SKILL.md in there. This is what Claude reads at the start of every chat to know how to behave.
3. Look for the knowledge base section on the right side of the project — there's a "+" button. Click it and upload the three files one at a time: results-analysis.md, deep-dive.md, and quick-analysis.md. These are the detailed playbooks Claude pulls from depending on what type of announcement you drop in.
4. That's it. From now on, any chat inside that project automatically has the full framework active — just paste an ASX announcement or drop in a PDF and it fires.

Cheers again

JM

That is very cool @JohnnyM

It also made me laugh.Glad it's not just me asking AI lots of newbie coding questions! (Until recently I had no idea what a repo was either lol). Amazing how quick you can learn with an AI by your side. And actually doing it --writing code and running it, well, it is absolutely the best way to learn.

I actually gave up on OpenClaw (aka Moltbook, aka clawbot). I just found it way too unstable.

But now I'm building an agent using cloudflare's agent SDK. MUCH more robust. Progress is slow, but I am moving forward. One stupid mistake at a time.

Can I ask how you built Bob? How do you handle memory and tooling (do you use MCP servers?), do you use an API to connect an AI to custom front end? What GitHub project did you use?

I'm heading deeper and deeper down this rabbit hole, and loving it. It's certainly made me very bullish on the future of this tech.

Thats awesome, any useful links etc would be handy to video guides or to particular tools you used to make it, got me pretty interested in having a crack if time allows..

This looks awesome. Well done

Double thumbs up @JohnnyM , you're an inspiration to this AI luddite!

Just this week I have started to get serious about using AI consistently and learning more - you have just provided a near term project and learning goal I feel I can achieve and, in the process, significantly expand my AI capabilities.

Cheers.

For all those who still didn't know (including me) 'what is repo' see below ;D

EDIT: in this case Repo refers to GitHub Repositories which is where the program sits and runs from. Not a Repo as in Repurchase Agreement, NOT WHAT I PUT BELOW, haha. That's what I get for having to Google a term myself.

A repo is a short‑term, collateralised loan dressed up as a sale and later repurchase of securities.

Plain‑English definition

In a repurchase agreement (“repo”):

• One party sells a security (usually government bonds) to another party for cash today.
• At the same time, they promise to buy back the same security later at a slightly higher price.
• The difference between the sale and repurchase price is effectively the interest on the cash loan (the repo rate).

Economically, the “seller” is borrowing cash and posting the security as collateral; the “buyer” is lending cash and holding the security until it’s repurchased.

Why repos matter

• They’re a core money‑market funding tool for banks, dealers, hedge funds, etc.
• Central banks use repo and reverse‑repo operations to add or drain liquidity and help keep overnight rates near their policy target.
• Because they’re secured by high‑quality collateral, repo rates are usually lower than unsecured interbank rates.

So, when you hear “repo”, think: very short‑term secured funding using securities as collateral, structured as ‘sell now, buy back later’.

Thanks for the interest in my post on Bob the Bot.. I thought the quickest way to share what I did was to make a quick video.

Felix and I explain what we've been working on here.

To recap the tools I used.

Python

Github

An API key from Open AI (which you will need to pay for usage)

An API key with Google Cloud and App Password for Gmail (for now it's free)

Things like Playwrite App to get around the ASX Gate, Chatty will write the code to call that function in Python.

Hope that helps

Cheers

JM+Felix

PS @tomsmithidg in this case Repo refers to GitHub Repositories which is where the program sits and runs from. Not a Repo as in Repurchase Agreement.

@Schwerms if you use windows, there is an icon for copilot, you can ask it all sorts of questions and you can give it jobs to do. I've had it write programs for me, which create python scripts you can run under windows. One of my scripts runs a scraper of the ASX website for all the shares I own, down load them and put them in directories. I can then use copilot to do deep dives into those shares, I go back five years, in my downloads, this then enables me to ask copilot to work out are they meeting targets, are the board doing what they say they are doing and check against track records. Amazingly powerful stuff, all for free. I'm so happy with how it is all going that I'm building a standalone AI inhouse system as I don't like the world knowing my shit. The copilot has designed a system for me, which I'm waiting on all the bits to arrive. All very exciting.

Probably using the ASX website for scraping, is they change the format and how the data is displayed regularly, so you need a script which can adjust.

@JohnnyM, to say this is inspirational is a complete understatement - many thanks for sharing!

I use Chat, a bit of python and Github for my home automation programming, but always for a small, defined specific piece of code. Nothing as big a workflow as this. The thoughts coming to my head:

1. Bloody hell .... this is powerful.

2. In virtually ALL the AI-related articles and video's I have viewed/read, the one advice that has stood out is "give the damn thing a go". Your video has given me a good and welcome kick in the nuts to say "give it a go, but think BIGGER, get it to do MORE"

3. I need to remove the mental block of "physically doing myself" and let AI do it for me - this, I think, is going to be my greatest inertia. I think when I do, so I need to learn how to think when I read ...

4. Need to think more closely on my current workflow, and how AI can speed it up for me. There are lots of opportunities to do that.

5. Get on with it ... NOW

Thanks again mate, very much appreciated!

Thankyou for doing this

Thanks @Foxlowe

Haha, thanks @JohnnyM :D